(Chest. 2002;122:84-91.)
© 2002
American College of Chest Physicians
Effect of a Nonrebreathing Exhalation Valve on Long-term Nasal Ventilation Using a Bilevel Device*
Nicholas S. Hill, MD, FCCP;
Carol Carlisle and
Naomi R. Kramer, MD, FCCP
* From the Rhode Island Hospital and Brown University School of Medicine, Providence, RI.
Correspondence to: Nicholas S. Hill, MD, FCCP, Pulmonary Division, Tufts-New England Medical Center, 750 Washington St, #257, Boston, MA 02111
 |
Abstract
|
|---|
Study objective: To determine whether an exhalation valve designed to minimize rebreathing improves daytime or nocturnal gas exchange or improves symptoms compared with a traditional valve during nocturnal nasal ventilation delivered using a bilevel pressure ventilation device.
Design: Prospective direct comparison trial with each patient sequentially using both valves, during a 2-week run-in period with a traditional valve, a 2-week trial with the nonrebreathing valve, and a 2-week washout period with the traditional valve.
Setting: Outpatient pulmonary function laboratory and home nocturnal monitoring.
Patients: Seven patients who received long-term (> 1 year) nocturnal nasal bilevel pressure ventilation with an expiratory pressure of 
4 cm H2O.
Intervention: Symptoms, pulmonary function, and arterial blood gas levels were assessed at each of three daytime sessions after the sequential 2-week periods using the different valves. Nocturnal studies used a multichannel recorder that measured heart rate, chest wall impedance, nasal airflow, and oximetry. End-tidal PCO2 (PetCO2) from the mask and transcutaneous PCO2 (PtcCO2) were also monitored nocturnally.
Results: Seven patients with a variety of neuromuscular, chest wall, and obstructive defects were enrolled. No mean differences in daytime arterial blood gas levels, pulmonary functions, nocturnal vital signs or oximetry, or PtcCO2 were apparent regardless of the exhalation valve used. The multichannel recording was indicative of an air leak at least one third of the time, and the PetCO2 tracing detected a blunted signal or no signal from the mask during the majority of the recording time.
Conclusion: The use of an exhalation valve designed to minimize rebreathing did not improve daytime or nocturnal gas exchange or symptoms in patients receiving long-term nasal bilevel pressure ventilation in comparison with a traditional exhalation valve, most likely because of air leakage and escape of CO2 via other routes.
Key Words: exhalation valves • nasal ventilation • noninvasive ventilation • rebreathing
|
Introduction
|
|---|
Noninvasive positive pressure ventilation (NPPV) has assumed an important role in the management of both acute and chronic forms of respiratory failure.1
2
Although many different positive-pressure ventilator types have been used to successfully administer NPPV, the use of so-called "bilevel" respiratory assist devices as ventilators to deliver NPPV has increased.3
4
5
These are blower-based ventilator systems that alternate between preset higher inspiratory and lower expiratory pressures as determined by patient triggering or a timer. In the outpatient setting for long-term intermittent NPPV, these devices offer a number of advantages over volume-limited portable ventilators, including greater portability, quiet operation, the ability to compensate for leaks, and lower cost.3
Bilevel devices also have several disadvantages compared with volume-limited portable ventilators, including less pressure-generating capacity, greater rates of energy consumption, and the lack of internal batteries. In addition, these devices use a single ventilator tube for both inspiration and expiration, which has the potential to promote rebreathing. Ferguson and Gilmartin6
and Lofaso et al7
8
demonstrated that substantial CO2 rebreathing occurs when a traditional exhalation valve (Whisper Swivel; Respironics; Murrysville, PA), a fixed valve with three narrow slits to permit exhalation of CO2, is used with a bilevel pressure ventilation device (BiPAP; Respironics). Rebreathing is greatest at low expiratory pressures (ie, 4 cm H2O), when bias flow during exhalation is lowest and the rebreathed gas can amount to 55% of each tidal volume.7
Lofaso et al8
demonstrated that a nonrebreathing valve eliminates rebreathing, but at the expense of increased expiratory resistance. Ferguson and Gilmartin6
found that the Plateau valve (Plateau; Respironics) also eliminates rebreathing. This valve, designed to minimize rebreathing by incorporating a variable resistor that permits more air to escape at low expiratory pressures than with the traditional valve, would not be expected to add to expiratory resistance.
In the present trial, we studied a small series of patients with chronic respiratory failure who had received treatment for at least 1 year with intermittent NPPV using bilevel pressure ventilation. These patients were using the traditional valve (Whisper Swivel) at low expiratory pressures ( 4 cm H2O), a condition that, based on the findings of Ferguson and Gilmartin,6
would be expected to promote rebreathing. To determine whether a valve designed to minimize rebreathing would further improve daytime and nocturnal gas exchange and thereby symptoms, we switched the patients to a nonrebreathing valve (Plateau) for a 2-week period, and then back to the traditional valve.
|
Materials and Methods
|
|---|
Patients
Patients were recruited from a group monitored by the Center for Noninvasive and Home Mechanical Ventilation at Rhode Island Hospital. To be eligible, patients had to have chronic respiratory failure stabilized with a bilevel pressure ventilation device using a nasal mask and a traditional valve. Expiratory pressure could not be > 4 cm H2O, and the clinical course and daytime gas exchange had to have been stable using NPPV for at least the previous year ( 6 mm Hg change in PaCO2, oxygen saturation 
90% on room air). Patients were ineligible if they had evidence of a respiratory infection, including increased cough, sputum production, or a febrile illness within the previous month. They were also excluded if they received NPPV for > 12 h per 24-h period. The Institutional Review Board at Rhode Island Hospital approved the study, and all patients gave written informed consent.
Study Design
After an initial 2-week run-in period using the traditional valve, patients underwent daytime and nocturnal studies. The next morning, the nonrebreathing valve replaced the traditional valve. After 2 weeks of using the nonrebreathing valve without any other alterations in ventilator settings, daytime and nocturnal studies were repeated. On the following day, patients resumed use of the traditional valve for an additional 2 weeks, after which daytime and nocturnal studies were repeated for a third time. The two valves are shown in Figure 1
.
View larger version (83K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1.. Photograph of the two exhalation valves used in the study, attached to standard nasal masks and standard tubing. Left: the arrow points to the three slits in the traditional valve, which permit exhalation of carbon dioxide. Right: the arrow points to the nonrebreathing valve, which contains a silicon diaphragm to permit proportionately more exhaled gas to escape at low expiratory pressures compared with high expiratory pressures.
|
|
Daytime Studies
Patients kept a daily symptom log throughout the course of the study that included visual analog scales of dyspnea, morning headache, and energy level. At the three daytime visits in the pulmonary function laboratory, they also rated their satisfaction with the ventilator system, including noise level, appearance, ease of inspiratory triggering, comfort of airflow, and overall satisfaction. All ratings were on a visual analog scale, with 1 representing the least or worst and 10 representing the most or best, as previously described.9
10
At these visits, patients’ somnolence was also assessed using the Epworth Sleepiness Scale.11
At the first daytime visit, exhaled tidal volume during bilevel pressure ventilation was measured during awake breathing using an in-line pneumotachograph (Ohmeda 4300; Ohmeda; Englewood, CA) to compare delivered tidal volumes during use of each valve. The pneumotachograph was placed between the mask and the exhalation valve, and leaks were minimized by assuring a tight air seal and nasal breathing. Exhaled tidal volume was determined from an average of 10 breaths using the patient’s own ventilator system at the usual settings. At each of the three daytime visits, vital signs were also obtained and the forced spirogram was measured according to American Thoracic Society standards,12
using a portable spirometry system (Renaissance; Puritan Bennett; Galena, KS). The best of three vital capacity maneuvers was selected. Arterial blood gas levels were obtained from a radial artery while patients sat comfortably breathing room air, and were analyzed using a Radiometer analyzer (Copenhagen, Denmark).
Nocturnal Studies
An EdenTrace II multichannel recorder was used for the three home nighttime studies (Tyco; St. Louis MO), which included tracings of oxygen saturation, heart rate, chest wall motion using piezocrystal electrodes, and airflow via nasal thermistry. End-tidal PCO2 (PetCO2) was recorded continuously, using a Nellcor PCO2 monitor (Nellcor Puritan Bennett; Carlsbad, CA) sampling via a catheter connected tightly to a nipple on the mask, and transcutaneous PCO2 (PtcCO2) was monitored using a TINA device (Radiometer). Monitoring was performed in the patient’s home, initiated by one of the investigators, starting at the hour of sleep and continuing until awakening in the morning.
Analysis of the nocturnal recordings included average heart rate, respiratory rate, and oxygen saturation obtained by sampling every 10 min. Air leak was defined as a decrease of > 75% in the nasal thermistor signal with preservation of the chest wall excursion. Episodes with movement artifact were excluded, and to minimize the likelihood of classifying obstructive events as leak, leaks had to be at least 2 min in duration (most obstructive events are 10 to 50 s)13
and not associated with oxygen desaturation. The percentage of leak was determined as the total time spent leaking as a percentage of total monitoring time. As an approximate index of variability in the PetCO2 signal, the percentage of time that the signals were < 75% and < 50% of the peak end-tidal value was also determined.
Statistics
Average values were determined from patient diaries and from daytime studies and nocturnal recordings at each of the three study time points. Values at each time point were compared using analysis of variance for repeated measures and, if F ratios were significant, the Student Newman Keuls test was used to determine statistically significant differences between individual mean values. Data are presented as mean ± SE, and p < 0.05 was considered statistically significant.
|
Results
|
|---|
Patient Characteristics
Table 1
shows that all patients had severe pulmonary dysfunction. Four of the seven patients had severe restriction caused by chest wall deformity or neuromuscular disease, and three patients (patients 4, 5, and 7) had mixed restrictive and obstructive defects. Only patient 4 was significantly overweight as determined by a body mass index > 30. All patients had been receiving noninvasive ventilation for at least 2 years, and had experienced an average drop of 11 mm Hg in daytime PaCO2 and an increase of 18 mm Hg in daytime PaO2 after initiation of noninvasive ventilation (Table 1)
.
View this table:
[in this window]
[in a new window]
|
Table 1.. Patient Characteristics, Spirometry, and Arterial Blood Gas Levels (Room Air) Obtained Before and After Initiation of Nasal Ventilation*
|
|
Ventilator Settings
Table 2
shows settings for inspiratory and expiratory pressure and set backup rate (breaths/min). Patient 4 used the S mode and had no backup rate. Exhaled tidal volumes were similar during bilevel ventilation when awake patients used the nonrebreathing or the traditional valves (Table 2)
.
Daytime Studies
During each of the three daytime study sessions, mean (± SD) arterial blood gas values were unchanged from the first session (pH, 7.38 ± 0.02; PaO2, 53 ± 0.2; and PaCO2, 69 ± 4 mm Hg), to the second session after use of the nonrebreathing valve (pH, 7.39 ± 0.02; PaO2, 54 ± 3 mm Hg; and PaCO2, 68 ± 4 mm Hg), and to the third session (pH, 7.40 ± 0.02; PaO2, 53 ± 1 mm Hg; and PaCO2, 70 ± 4 mm Hg) [all p > 0.05]. Analysis of individual arterial blood gas values revealed decreases of 5 mm Hg in PaCO2 during nonrebreathing valve use compared with traditional valve use in patient 2 and patient 6 (Fig 2
). These values returned to the higher baseline levels on resumption of use of the traditional value. However, decreases in PaCO2 were not consistent among the other patients during nonrebreathing valve use, with four of the seven patients registering small increases (Fig 2)
.
View larger version (16K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2.. PaCO2 values for individual patients are displayed for the three daytime sessions, after the first period of use of the traditional valve (Whisper-Swivel, left), after 2 weeks of use of the nonrebreathing valve (Plateau, middle), and 2 weeks after switching back to the traditional valve (right). The numbers above the lines correspond to the individual patients. The larger solid diamond connected by the darker interrupted line represents the mean values for the group. Differences between mean values were not statistically significant.
|
|
Paralleling the mean arterial blood gas values, symptoms scores, including energy level, dyspnea, and sleepiness, remained unchanged when patients switched from the traditional valve to the nonrebreathing valve and back (Table 3
). In addition, FVC, FEV1, and maximal expiratory pressure remained unchanged during the three daytime testing sessions. Maximal inspiratory pressure was significantly lower after 2 weeks of use of the nonrebreathing valve than after the initial period of use of the traditional valve use, but the clinical significance of this difference is dubious because there was no return to the initial value after resumption of use of the traditional valve (Table 3)
.
Nocturnal Measures
Mean nocturnal values for heart and respiratory rate, as well as oxygen saturation and PtcCO2, did not differ between the study periods (Table 4
). PetCO2 exhibited considerable variability during the recording and did not parallel changes in PtcCO2. As shown in Table 4
, the PetCO2 signal was < 75% of the peak value 60% of the time, and < 50% of the peak value approximately one fourth of the time. Figure 3
shows the variability in the PetCO2 signal during overnight monitoring, as sampled every 10 min. Long periods (> 1 h in some patients) transpired, during which time the PetCO2 signal was undetectable, corresponding to episodes of leaking on the multichannel recording. Figure 4 shows examples of air leaking (lower panel) and the absence of leaking (upper panel) during use of the multichannel recorder. Air leak, as we defined it, occurred during approximately one third of the recording time (Table 4)
.
View larger version (19K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3.. PCO2 values recorded from the nasal mask and corresponding PtcCO2 are shown for patient 3 as sampled every 10 min. PetCO2 was undetectable for prolonged periods during nocturnal monitoring, and there were no corresponding changes in the PtcCO2 tracing.
|
|
View larger version (140K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4.. Sample tracings from patient 3 of the multichannel recording used for nocturnal monitoring. Four channels are shown, displaying heart rate at the top, chest wall impedance next, airflow as determined by nasal thermistry next, and oxygen saturation using a pulse oximeter at the bottom. The upper panel exemplifies a period when no leak was apparent. Regular chest wall motions are detected in the impedance tracing, and synchronous, easily detectable signals are seen on the airflow recording. The lower panel exemplifies leak. Regular chest wall motions are again detectable, but airflow barely registers, even though oxygen saturation in this patient, while breathing room air, was well maintained. Both tracings were obtained from within a 1-h span on the same night.
|
|
Subjective Ratings of Different Exhalation Valves
Patients found the nonrebreathing valve to be noisier and less attractive in appearance than the traditional valve, the latter related to the greater bulkiness of the nonrebreathing valve (Table 5
). They also rated both valves as comparable in terms of ease of triggering and comfort of airflow. Although the traditional valve garnered a slightly higher overall rating, the difference was not statistically significant.
|
Discussion
|
|---|
Our study found no substantial differences in the performance of two different exhalation valves commonly used for noninvasive ventilation administered with a bilevel pressure ventilation device, as assessed by daytime and nocturnal gas exchange and symptoms. Although our findings are contrary to our initial hypothesis, we should consider a number of limitations of our study design before rejecting the hypothesis. First, we did not quantify the amount of rebreathing our patients experienced. In their lung model study, Ferguson and Gilmartin6
demonstrated substantial rebreathing during bilevel pressure ventilation with the traditional valve unless expiratory pressure was raised > 4 cm H2O. Similarly, Lofaso et al7
showed that rebreathing amounted to 55% of the ventilator breath when the bilevel pressure ventilation device was set at inspiratory and expiratory pressures of 15 cm H2O and 2 cm H2O, respectively. Ventilator settings for our patients averaged 15 cm H2O and 3 cm H2O for inspiratory and expiratory pressures, respectively, well within the range that should have contributed to substantial rebreathing during use of the traditional valve.
Another limitation is that our study period of 2 weeks may have been too short to detect differences in daytime gas exchange. However, differences in nocturnal gas exchange should have been readily apparent had there been substantial differences in the performance of the two valves. Also, by focusing on average responses in blood gas levels and symptoms, we may have missed individual patients whose nocturnal gas exchange was improved by the use of the nonrebreathing valve, even when most patients did not have favorable responses. The substantial ( 5 mm Hg) drop in daytime PaCO2 during use of the nonrebreathing valve observed in patient 2 and patient 6 is consistent with this view. However, improvements in nocturnal PtcCO2 and oxygen saturation did not accompany the improvements in PaCO2 in these patients (data not shown), and the apparent improvements in gas exchange are more likely to reflect experimental variability than any real gains. Further, our visual analog scale has not been fully validated and may have been inadequate to detect differences in some of our measures, although it has detected significant differences in dyspnea, comfort, and ventilator characteristics in prior studies.9
10
If the different exhalation valves functioned as described in the prior studies,6
7
8
and the nonrebreathing valve reduced rebreathing in our patients compared with the traditional valve, several possibilities may explain the lack of effect on gas exchange. First, patients may have altered their breathing patterns during use of nocturnal bilevel pressure ventilation to compensate for differences in rebreathing. This pattern of response was observed in the acute setting by Lofaso et al,8
whose patients increased their tidal volumes when using the traditional valve. However, this explanation seems unlikely because we measured tidal volumes at the patients’ usual ventilator settings and found no differences between the traditional and nonrebreathing valves. Although these tidal volumes were measured during awake breathing, and it remains possible that breathing patterns differed during nocturnal breathing, most of our patients had neuromuscular disease and permitted the ventilator to control their breathing nocturnally. Nocturnal respiratory rates were not significantly changed throughout, and it seems unlikely that these restricted patients would have increased tidal volumes without altering respiratory rate. Also, there were no changes in nocturnal vital signs that might have reflected differences in breathing effort.
A more likely explanation for the similar performances of the exhalation valves is the frequent occurrence of air leaking that provides an alternative route for the exhalation of carbon dioxide. We14
and others15
have demonstrated that mouth leaking during nasal ventilation occurs very commonly in patients receiving NPPV, occupying the majority of sleep, and the entirety of certain sleep stages such as slow-wave sleep, if very sensitive detection techniques such as continuous video monitoring are used. In the current study, leaking as we defined it was apparent approximately one third of the recording time, less than in our prior study.14
However, the leak estimation we obtained from the multichannel recorder was used only as an indication of substantial air leaking but not as a quantitative measure and has not been validated in comparison to full polysomnography. Episodes of partial leaking, when some air leaked through the mouth or under the mask, were undoubtedly missed. Furthermore, PetCO2 recording from the mask during nocturnal studies revealed a blunted signal for the majority of sleep, consistent with the escape of carbon dioxide via an alternative route.
Thus, it is likely that a substantial amount of carbon dioxide was expelled through the mouth or via other routes besides the exhalation valve during nocturnal nasal ventilation, reducing the potential consequences of rebreathing. This would seem to be the most likely explanation for the lack of improvement in nocturnal or daytime gas exchange when the nonrebreathing valve was substituted for the traditional valve.
Although our findings indicate that the potential advantages of using an exhalation valve designed to minimize rebreathing during bilevel pressure ventilation are not realized in field testing, a number of caveats should be borne in mind. First, ours were patients with chronic respiratory failure receiving NPPV only nocturnally. These findings may not apply to patients in the acute setting or receiving NPPV continuously. Second, our patients used nasal masks, and patients using oronasal masks might be more prone to rebreathing with a traditional valve because of reduced air leaking through the mouth. Finally, ours is a small study, and it is conceivable that some individual patients, particularly those with marked carbon dioxide retention and minimal air leaking through the mouth, could still be benefited by use of the nonrebreathing valve. Nonetheless, by virtue of the crossover design, the power was at 85% to detect a difference (at 
= 0.05) of 2.5 mm Hg in PaCO2.
It might be argued that the nonrebreathing valve should be used routinely, despite the findings of our study, because of the theoretical reduction in rebreathing. However, patients did find the nonrebreathing valve noisier and less appealing in appearance, and they tended to prefer the traditional valve. Considering that these attributes of the nonrebreathing valve might adversely affect tolerance of NPPV in some patients and add to expense, it is difficult to support this view.
Based on the findings of Ferguson and Gilmartin6
and Lofaso et al,7
8
that increases in expiratory positive airway pressure > 4 cm H2O reduce the amount of rebreathing, some16
have recommended that expiratory pressure be kept at a minimum level, such as 5 cm H2O. Considering that our patients used expiratory pressures ranging from 2 to 4 cm H2O, however, this practice appears to be unnecessary unless it can be demonstrated that such pressures are useful in eliminating hypopneas or apneas. Routine administration of expiratory pressures greater than the minimal level necessitates using greater inspiratory pressure to maintain a given level of pressure support, and this may adversely affect tolerance.
We conclude that substituting an exhalation valve designed to minimize rebreathing for an exhalation valve known to be associated with rebreathing does not offer discernible advantages for patients with chronic respiratory failure who receive long-term nocturnal nasal ventilation. Further studies will be necessary to determine whether exhalation valves designed to minimize rebreathing offer measurable advantages to patients in other clinical settings, or whether other approaches to eliminating rebreathing, such as creating fixed leaks in the mask or using separate inspiratory and expiratory limbs in the ventilator circuit, might be more efficacious.
|
Acknowledgements
|
|---|
We thank Jan Hayden for assistance in preparation of this article; Respironics, Inc., for providing a research grant to support the study; and, most of all, our patients, for their willingness to participate.
|
Footnotes
|
|---|
Abbreviations: NPPV = noninvasive positive pressure ventilation; PetCO2 = end-tidal PCO2; PtcCO2 = transcutaneous PCO2
This study was supported by a research grant from Respironics, Inc., Pittsburgh, PA.
Received for publication July 2, 2001.
Accepted for publication January 7, 2002.
|
References
|
|---|
- Mehta, S, Hill, NS (2001) Noninvasive ventilation: state of the art. Am J Respir Crit Care Med 163,540-577[Free Full Text]
- Meduri, GU (1996) Noninvasive positive-pressure ventilation in patients with acute respiratory failure. Clin Chest Med 17,513-553[CrossRef][ISI][Medline]
- Strumpf, DA, Carlisle, CC, Millman, RP, et al (1990) An evaluation of the Respironics BiPAP bi-level CPAP device for delivery of assisted ventilation. Respir Care 35,415-422
- Hill, NS, Eveloff, SE, Carlisle, CC, et al (1992) Efficacy of nocturnal nasal ventilation in patients with restrictive thoracic disease. Am Rev Respir Dis 101,516-521
- Meecham Jones, DJ, Paul, EA, Jones, PW (1995) Nasal pressure support ventilation plus oxygen compared with oxygen therapy alone in hypercapnic COPD. Am J Respir Crit Care Med 152,538-544[Abstract]
- Ferguson, GT, Gilmartin, M (1995) CO2 rebreathing during BiPAP ventilatory assistance. Am J Respir Crit Care Med 151,1126-1135[Abstract]
- Lofaso, F, Brochard, L, Touchard, D, et al (1995) Evaluation of carbon dioxide rebreathing during pressure support ventilation with airway management system (BiPAP) devices. Chest 108,772-778[Abstract/Free Full Text]
- Lofaso, F, Brochard, L, Thierry, H, et al (1996) Home versus intensive care pressure support devices: experimental and clinical comparison. Am J Respir Crit Care Med 153,1591-1599[Abstract]
- Kramer, N, Meyer, TJ, Meharg, J, et al (1995) Randomized, prospective trial of noninvasive positive pressure ventilation in acute respiratory failure. Am J Respir Crit Care Med 151,1799-1806[Abstract]
- Hill, NS, Mehta, S, Carlisle, CC, et al (1996) Evaluation of the Puritan-Bennett 335 portable pressure support ventilator: comparison with the Respironics BiPAP S/T. Respir Care 41,885-894
- Johns, MW (1991) A new method for measuring daytime sleepiness: the Epworth Sleepiness Scale. Sleep 14,540-545[ISI][Medline]
- Committee on Proficiency Standards for Clinical Pulmonary Function Laboratories: standardization of spirometry. Am J Respir Crit Care Med 1995;152,1107-1136[ISI][Medline]
- Littner, MR, Shepard, JW (1999) Recommendations for research into measurement and classification of sleep disordered breathing: gazing into the crystal ball; AASM Task Force. Sleep 22,667-689[ISI][Medline]
- Meyer, TJ, Pressman, MR, Benditt, J, et al (1997) Air leaking through the mouth during nocturnal nasal ventilation: effect on sleep quality. Sleep 20,561-569[ISI][Medline]
- Bach, JR, Dominique, R, Leger, P, et al (1995) Sleep fragmentation in kyphoscoliotic individuals with alveolar hypoventilation treated by NPPV. Chest 107,1552-1558[Abstract/Free Full Text]
- Hill, NS, Hess, D (2001) Initiation of noninvasive positive pressure. Hill, NS eds. Noninvasive positive pressure ventilation: principles and applications ,27-40 Futura Publishing Company Armonk, NY.
TOP
Abstract
Introduction
